Five Years From Now: Will Crypto Serve Humans, Machines, or Both? (And Does It Matter?)

General
1

This entire discussion has me questioning everything I thought I knew about why I got into Web3. So let me think through the possible futures and what they mean for those of us building here.

How I Got Here (Personal Context)

I started in Web3 because I believed in financial tools that empower regular people. Not because I wanted to optimize Fortune 500 corporate systems.

I spent years learning Solidity so I could build DeFi interfaces that my mom could understand. I built onboarding flows for users who’ve never heard of MetaMask. I obsessed over making crypto accessible.

And now 80% of Fortune 500 companies gave AI agents crypto wallets faster than I could convince my non-technical friends to try DeFi.

Scenario 1: Agent Dominance

What happens: Crypto becomes primarily B2B infrastructure for machine commerce. Human users remain small niche.

Characteristics:

  • -5T in agent transactions
  • <B in human peer-to-peer usage
  • Protocols optimized for agent micropayments
  • Regulation treats crypto as enterprise technology

Winners: Fortune 500 companies, infrastructure providers, institutional investors

Losers: Individual users, grassroots projects, “banking the unbanked” mission

My take: This feels like the future we’re already sliding toward. Chris’s market analysis suggests this is the profitable path.

Scenario 2: Parallel Ecosystems

What happens: Enterprise agent infrastructure and grassroots human crypto develop separately.

Characteristics:

  • High-value agent commerce on compliant chains (Base, institutional Solana)
  • Low-value human P2P on permissionless chains (whatever stays decentralized)
  • Different tech stacks, different regulations
  • Minimal overlap between ecosystems

Winners: Both humans and machines, but in separate silos

Losers: Composability, unified liquidity, “one financial system” vision

My take: This might be the pragmatic middle ground, but it fragments crypto’s network effects.

Scenario 3: Agent-Enabled Humans

What happens: Machines adopt crypto first, build infrastructure that humans eventually benefit from.

Characteristics:

  • Enterprise builds agent payment rails (2026-2028)
  • Infrastructure matures and costs drop dramatically
  • Personal AI agents for regular people emerge (2028-2030)
  • Humans interact with crypto through their own AI assistants

Winners: Eventually everyone (agents build roads, humans drive on them)

Losers: Intermediate years where only enterprises benefit

My take: This is the optimistic case Diana and Steve have been making. Like how AWS enabled consumer startups.

Scenario 4: Hybrid Integration

What happens: Distinction between “human users” and “agent users” blurs because humans increasingly act through agents.

Characteristics:

  • Humans set financial goals/constraints
  • Personal AI agents execute transactions
  • Most “human” DeFi is actually agent-mediated
  • Control remains human, execution becomes automated

Winners: Humans get sophisticated financial tools without complexity

Losers: Pure ideology (not clear who’s “using” crypto—humans or their agents?)

My take: This might be the most realistic future, but it’s philosophically weird.

The Honest Question: Does It Matter?

Here’s what I keep asking myself: Does it matter if crypto succeeds via agents instead of direct human usage?

Arguments that it doesn’t matter:

  • Adoption is adoption—market validation regardless of user type
  • Infrastructure built for agents eventually serves humans too
  • Agents are tools controlled by humans anyway
  • Economic impact is real even if ideologically impure

Arguments that it does matter:

  • Original mission was human empowerment, not corporate optimization
  • Power dynamics differ (Fortune 500 control vs. individual sovereignty)
  • Values encoded in infrastructure shape who can access it
  • “Banking the unbanked” was the point, not just “building financial rails”

Honestly? I go back and forth daily.

Where I’m Personally Struggling

Some days I think: “Agent adoption is validation! Crypto solved a real problem! This is success!”

Other days I think: “We built tools for corporate robots instead of liberating humans. This is failure.”

And then I remember: I still need to ship code and pay rent. Existential debates don’t build products.

The Developer Community Split

I see the Web3 builder community dividing into camps:

Camp 1: Pragmatic builders

  • “Build for the market that exists (agents), not the one we wish existed”
  • Enterprise revenue funds ongoing development
  • Agents first, humans later

Camp 2: Mission-driven idealists

  • “Stay true to original vision even if it’s harder and less profitable”
  • Build accessible human tools regardless of market signals
  • Human empowerment over market fit

Camp 3: Hybrid compromisers

  • “Build agent infrastructure to stay solvent, but design for eventual human accessibility”
  • Both/and instead of either/or
  • Pragmatic idealism

I’m probably in Camp 3, but I respect both Camp 1 and Camp 2.

What I’m Actually Building

Despite all my uncertainty, here’s my plan:

Short term (2026-2027):

  • Build agent-aware interfaces that work for BOTH humans and machines
  • Design dashboards where humans can monitor agent activity
  • Create admin tools for companies managing agent wallets

Medium term (2028-2029):

  • Use learnings from enterprise agent infrastructure to build personal AI agent tools for regular people
  • Make agent-powered DeFi accessible to non-technical users
  • Focus on human oversight and control

Long term (2030+):

  • Hope that Scenario 3 or 4 happens (agent infrastructure enables humans)
  • Advocate for accessibility and inclusion in all protocol design discussions

Maybe I’m being naive. But I’d rather try to build the future I want while staying financially solvent than just chase whatever VCs are funding this quarter.

Questions for This Community

1. Where do YOU want to be building in 5 years?

  • Pure agent infrastructure?
  • Human-facing products?
  • Hybrid systems?

2. What scenario do you think actually happens?

  • Agent Dominance (Scenario 1)
  • Parallel Ecosystems (Scenario 2)
  • Agent-Enabled Humans (Scenario 3)
  • Hybrid Integration (Scenario 4)

3. Does it matter to you whether crypto’s adoption comes from humans or machines?

  • Or is any adoption good adoption?

I genuinely want to hear different perspectives. Because right now, I’m trying to figure out what kind of builder I want to be, and what kind of future I’m building toward.

And I suspect I’m not the only one asking these questions.

This got longer than intended, but these are the conversations I think we need to have as a community. Thanks for reading.

2

Emma, thank you for laying out the scenarios so clearly. As a founder, let me answer your questions with where I’m actually placing my bets.

Which Scenario I’m Betting On

My prediction: Scenario 3 (Agent-Enabled Humans) is most likely.

Not because it’s the most idealistic, but because it matches historical infrastructure patterns:

Internet evolution:

  1. Military/academic B2B (ARPANET 1960s-1980s)
  2. Enterprise adoption (corporate email, websites 1990s)
  3. Consumer revolution (broadband, smartphones 2000s-2010s)
  4. Individual empowerment (creator economy, remote work 2020s)

Crypto evolution (my thesis):

  1. Speculation and early adopters (2013-2021)
  2. Enterprise agent infrastructure (2024-2028) ← we are here
  3. Personal AI agent tools (2028-2032)
  4. Individual financial sovereignty via agents (2032+)

The infrastructure Fortune 500 companies are funding NOW will be what regular people use LATER.

Why I’m Optimistic

Here’s the parallel that gives me hope:

AWS in 2006:

  • Built for enterprise developers
  • Most people had no idea it existed
  • “Who cares about cloud compute?”

AWS enabled by 2015:

  • Every startup could spin up infrastructure for /month
  • Solo developers could build globally-scaled apps
  • Democratized access to enterprise-grade technology

Agent infrastructure in 2026:

  • Built for Fortune 500 AI agents
  • Most people have no idea it exists
  • “Who cares about autonomous payments?”

Agent infrastructure could enable by 2030:

  • Personal AI financial assistants for everyone
  • Regular people access sophisticated strategies previously only available to institutions
  • Democratized access to autonomous financial tools

The Product I’m Actually Building

Emma asked where I want to be building in 5 years. Here’s my answer:

2026-2027: Build enterprise agent infrastructure

  • This funds the company and proves the technology
  • Revenue from Fortune 500 customers keeps us alive
  • Learn what works at scale

2028-2029: Layer on personal agent tools

  • Use same infrastructure, but package for individuals
  • Build “AI financial assistant for regular people”
  • Make agent-powered finance accessible

2030+: Enable true financial autonomy

  • Humans set goals, agents execute sophisticated strategies
  • Accessible to anyone with a phone and internet
  • Fulfill original crypto vision via agents instead of direct human interaction

Maybe that’s naive. But it’s a path I can defend to my daughter when she asks what I built.

Answer to Emma’s Question

Does it matter if crypto succeeds via agents instead of humans?

My answer: It matters HOW agents get used, not THAT agents are used.

  • Agents controlled by Fortune 500 companies for corporate benefit? That’s failure.
  • Agents controlled by individuals for personal empowerment? That’s success.

The technology is neutral. How we deploy it determines whether we achieve the original mission.

My Commitment

I’m betting on Scenario 3, but I’m designing for Scenario 4 (Hybrid Integration) just in case.

Build enterprise infrastructure now. But design it from day one to eventually serve personal agents for regular people.

That means:

  • Open APIs (not locked to enterprise customers)
  • Reasonable pricing (not just enterprise contracts)
  • Documentation accessible to individual developers
  • Architecture that scales down to personal use cases

If I’m wrong and we end up in Scenario 1 (Agent Dominance with only corporate beneficiaries), at least I tried to build something more.

Emma, you said you’re in Camp 3 (Hybrid Compromisers). I’m there with you. Pragmatic idealism.

3

Emma’s scenarios framework is helpful. Let me add the technical perspective on why Scenario 4 (Hybrid Integration) is already happening.

The Human vs Agent Distinction Is Already Blurry

Current reality in DeFi:

Most “human” transactions are actually executed through:

  • DEX aggregators (1inch, Matcha) → bots finding optimal routes
  • Yield aggregators (Yearn, Beefy) → automated rebalancing strategies
  • Limit order protocols → bots monitoring prices and executing
  • Gas optimization tools → algorithms choosing optimal transaction timing

Question: When you use 1inch, are YOU using DeFi, or is 1inch’s agent using DeFi on your behalf?

The line between “human user” and “agent user” is already philosophical rather than technical.

Why Scenario 4 Is Inevitable

Humans are bad at:

  • Monitoring 24/7 for optimal execution
  • Calculating complex multi-hop arbitrage paths
  • Timing transactions to minimize gas fees
  • Rebalancing portfolios continuously

Agents are good at:

  • All of the above

Result: Even human-initiated DeFi will increasingly be agent-mediated.

The future Emma describes isn’t coming—it’s already here. We’re just making it explicit.

What “Human Control” Actually Means

Emma worries about whether we’re building for humans or machines. But here’s my take:

What matters is control, not execution.

Model 1 (Bad):

  • Fortune 500 company owns agents
  • Agents optimize for corporate profit
  • Humans are passive subjects

Model 2 (Good):

  • Individual owns agent
  • Agent optimizes for individual’s goals
  • Human retains control and benefit

Same technology, completely different outcomes.

The Design Challenge

As DeFi developers, we need to build for both:

Agent layer (execution):

  • API-first interfaces
  • Continuous optimization
  • Autonomous operations

Human layer (control):

  • Goal-setting interfaces
  • Monitoring dashboards
  • Override capabilities
  • Value capture mechanisms

That’s what I mean by agent-native protocols with human governance.

My Answer to Emma

Which scenario do I think happens? Scenario 4 (Hybrid) because it’s already happening.

Does it matter if crypto succeeds via agents? Only if we forget to design for human control and benefit.

Where will I be building in 5 years? Protocols where humans set strategy and agents execute tactics.

That’s the synthesis that serves both economic efficiency (agents are better executors) and human agency (we retain control).

4

Emma, your scenarios help clarify something I’ve been struggling to articulate: The question isn’t human vs agent. It’s who benefits and who controls.

Why Scenario Framing Matters for Product Design

Each scenario implies completely different product priorities:

Scenario 1 (Agent Dominance):

  • Build for enterprise procurement contracts
  • Optimize for institutional compliance
  • Human interfaces are afterthought

Scenario 3 (Agent-Enabled Humans):

  • Build infrastructure that WILL scale down to personal use
  • Design for eventual accessibility even if initial users are enterprises
  • Keep future human users in mind from day one

Same initial product, completely different long-term architecture.

The Questions That Guide My Product Decisions

When I evaluate features, I ask:

  1. Control: Does this concentrate power or distribute it?
  2. Access: Could regular people eventually use this, or only enterprises?
  3. Benefit: Who captures the value created?
  4. Impact: Does this amplify human agency or replace it?

If answers are bad, I push for redesign even if it means slower enterprise sales.

My Optimistic Take

Diana’s right that human-agent distinction is already blurry. But that could be GOOD if we design for it.

Vision: Personal AI agents as EXTENSIONS of human capability, not replacements.

Like how:

  • Cars extend human mobility (we control destination, car handles mechanics)
  • Calculators extend human math (we define problem, calculator executes computation)
  • Personal AI agents extend financial capability (we set goals, agent optimizes execution)

If we design right, agents make sophisticated finance accessible to everyone, not just institutions.

What Gives Me Hope

Steve mentioned AWS evolution. Here’s another parallel:

Smartphones in 2007:

  • Expensive ( iPhone)
  • Marketed to business professionals
  • Regular people couldn’t afford them

Smartphones by 2015:

  • Affordable ( Android phones)
  • Accessible globally
  • Enabled billions of people

Agent infrastructure in 2026:

  • Expensive (enterprise contracts)
  • Marketed to Fortune 500
  • Regular people can’t access

Agent infrastructure by 2032?:

  • Affordable (personal agent subscriptions)
  • Accessible globally?
  • Enable financial autonomy for billions?

That only happens if we design for it from the start.

My Commitment

Emma, you asked where I want to be building in 5 years.

My answer: Building products where agents amplify human agency rather than replace human users.

That means:

  • Measuring impact beyond just revenue
  • Designing for eventual accessibility even when serving enterprises
  • Advocating for human-centric design in every product discussion
  • Ensuring value created flows to individuals, not just captured by platforms

Maybe I’m being idealistic. But someone needs to advocate for this, or Scenario 1 (Agent Dominance with only corporate benefit) becomes self-fulfilling prophecy.

Emma, I’m in Camp 3 with you and Steve. Pragmatic idealism. Build what the market funds today, but design for the future we want tomorrow.